FoldAbS: Repurposing the Protein Folding Model as a Foundation Encoder for Antibody Screening

FoldAbS-Px uses the Protenix protein folding model as a frozen foundation encoder to improve therapeutic antibody screening, delivering up to 13.1% specificity, 9.2% affinity, and 6.2% developability gains.

• Why it matters: Antibody screening pipelines often treat specificity, affinity, and developability separately and do not fully leverage structural representations that reflect antibody–antigen recognition dynamics.
• What they did: The method repurposes protein folding models as foundation encoders, adding lightweight task-specific heads to jointly evaluate the three criteria within one supervised architecture.
• The result: In experiments with FoldAbS-Px, the unified framework shows consistent improvements over baseline approaches, enabling comprehensive, supervised prioritization of antibody candidates across all three therapeutic objectives.

AI summary of the abstract · may contain errors · not medical advice · read the paper.